1995

Hasemann, Jörg-Michael

Planning and monitoring in dynamic environments are the key activities for controlling a discrete event system such as a robot, manufacturing system, or chemical process plant. Typically, these systems are situated in a changing, only moderately predictable, environment. Correct and conflictfree plan execution demands both through plan validity and rigorous plan execution monitoring. Planning within real-world applications is often time critical and demands fast solutions. The thesis reviews and discusses traditional approaches towards planning and monitoring. New concepts for combined monitoring and planning are presented and formalized. The system presented features behaviour switching and complete monitoring of plan execution and plan validity using rigorous explicit representations of activities and mutual dependencies within plans. The key techniques used to model decision making and plan mdification are fuzzy logic and graph grammars. Fuzzy logic is proposed for guiding planning. Graph rewriting provides the framework for expanding the plan components.